10.18453/rosdok_id00005819 Leifert, Gundram Gundram Leifert Strauß, Tobias Tobias Strauß http://d-nb.info/gnd/1136294554 https://orcid.org/0000-0002-5344-4188 Labahn, Roger Roger Labahn http://d-nb.info/gnd/170999645 https://orcid.org/0000-0003-1901-9644 Wustlich, Welf Welf Wustlich University of Rostock, Institute of Mathematics 2012 510 Mathematics 2012 en https://purl.uni-rostock.de/rosdok/id00005819 urn:nbn:de:gbv:28-rosdok_id00005819-6 A recent approach for offline handwriting recognition is to use multidimensional recurrent neural networks (MDRNN) with connectionist temporal classification which has shown to yield very good results on several datasets. MDRNNs contain special units – multidimensional Long Short-Term Memory (MDLSTM) cells. These cells suffer from instability especially for higher dimensionality. We analyze the reasons for this effect and introduce several cells with better stability. We present a method to design stable multidimensional cells using the theory of linear shift invariant systems. The new cells are compared to MDLSTMs on the Arabic and French ICDAR datasets, where they yield better results.